Advised copying and restarting AI workflows to reset state and ensure proper functionality during iterative development.

Using VS Code IDE attachments to connect directly to AI project workflows, facilitating streamlined editing, file referencing, and debugging.

Tom Spencer and Cameron Rohn demonstrate setting up, refining, and adjusting prompts iteratively to ensure AI agents deliver the required level of det...Show more

Tom Spencer recommends leveraging task decomposition agents to build specialized sub-agent teams—like academic research experts—for modular and focuse...Show more

Tom Spencer highlights using meta-prompting techniques to rapidly generate and customize AI components within Claude Code projects for streamlined dev...Show more

Tom Spencer recommends defining a scripted 'init' phase for AI development workflows to automate environment setup and dependency loading for subagent...Show more

Cameron Rohn explains how slash commands in Claude Code act as shortcuts for configurations, settings, and MCP tools, enabling developers to quickly a...Show more

Tom Spencer demonstrates a live workflow using the Claude Code subagent to implement metaprompting, showing developers how to embed and trigger dynami...Show more

Implementing chaining techniques can improve how AI models handle sequential tasks and manage information flow in complex workflows

The number of AI tools loaded impacts context window consumption and can increase load times and resource usage, particularly on MCP servers handling ...Show more

Using AI to generate prompts accelerates development but requires well-defined best practices as guardrails to prevent the system going off the rails

Describing sub-agents using a consistent markdown structure with metadata facilitates clear documentation and parsing of the agent architecture

Using dedicated settings files introduces a new dimension of customization, letting you fine-tune agent behavior and policies without changing core co...Show more

Defining tools at a top-level agent and having sub-agents inherit them streamlines configuration and reduces redundancy across your agent hierarchy.

Leveraging JSON-based agent cards and an Agent-to-Agent (A2A) protocol provides a standardized abstraction for defining interactions and handoffs betw...Show more

Using a hierarchy of supervisors and sub-agents allows you to manage complex AI workflows by delegating responsibilities across specialized components...Show more

Creating MCP JSON files enables defining and managing agents with structured configurations for consistent behavior and easier updates

Teams can share MCP JSON configuration files and code setups among members using Claude to streamline collaboration on agent projects

Feeding relevant information into each agent enhances their understanding and performance by maintaining a consistent context across interactions

Effective context preservation relies on storing prompt information so each agent or LLM retains understanding based on previously fed data.